PPI use and dementia

a ​pharmacoepidemiological analysis, based on claims data, found an association between taking proton pump inhibitors (PPIs) and the risk of dementia in the elderly (see ppi and dementia jamaneurol2016 in dropbox, or doi:10.1001/jamaneurol.2015.4791). details:

--an observational database from 2004-2011 from the largest Germany health insurer insuring 1/3 of the population and 50% of the elderly, was used to assess inpatient and outpatient use of PPIs

--73679 participants, >=75yo and free of dementia at baseline. 2950 were on regular PPIs (mean age 83.8, 78% female), compared to 70,729 patients (mean age 83.0, 73.6% female)

--29510 patients received a diagnosis of dementia over the course of the study period (40% of the cohort)

--regular PPI use was defined as a prescription for a PPI each quarter of an 18-month interval during the study period

--results:

    --those on PPIs had a 44% increased risk of incident dementia [HR 1.44 (1.36-1.52), p<0.001]

        --for males, HR 1.52 (1.33-1.74)

        --for females, HR 1.42 (1.33-1.51)

    --other associations with dementia included: depression (HR 1.28), stroke (HR 1.37), and to a lesser extent, female sex (HR 1.15), diabetes (HR 1.05), polypharmacy (HR 1.16); all statistically significant

    ​--but, controlling for all these potentially confounding factors actually led to a higher association of dementia with PPI use [HR 1.66 (1.57-1.76)]

    --and, controlling for use of anticholinergics, itself a risk factor for dementia with its own HR of 1.80, did not change the association with PPIs (HR was still 1.44)

    --looking at those patients who only occasionally used PPIs, there was a lower but significant HR of 1.16 (1.13-1.19).

    ​--difference by type of PPI used: omeprazole had HR 1.51, pantoprazole HR 1.58, but somewhat higher with esomeprazole with HR 2.12

    ​--there was a slight decreasing effect with age: those 75-79 had HR 1.69, those 80-84 had HR 1.49 and those >84 had HR 1.32 (all significant)

--a prior study by the same group, AgeCoDe (the German Study on Aging, Cognition and Dementia in Primary Care Patients), included 3327 community-dwelling persons >= 75 yo with 18-month neuropsychological assessments similarly found incident dementia to be associated with PPI usage, and with a similar HR of 1.38 (1.04-1.83), and with somewhat more dementia in the group on esomeprazole. all PPI associations controlled for age, sex, educational level, apoE4, depression, diabetes, stroke, ischemic heart disease, and polypharmacy, 

so, this was a large computer-based study finding an association between PPI use and dementia; therefore it is not confirm a causal relationship. but a few points:

--there is biological plausibility:

    --mouse models find that PPIs increase the levels of b-amyloid in their brains, by affecting the enzymes b- and g-secretase; PPIs also could decrease the degradation of b-amyloid by lysosomes, which are pH-dependent, in microglia (and some PPIs cross the blood-brain barrier). also, PPIs bind to tau.

    --PPIs are associated with vitamin B12 deficiency through malabsorption, and vitamin B12 deficiency is associated with cognitive decline

--this study adds to the potential adverse effects of PPIs (see blogs below for some other problems with PPIs)

--of course, this study only finds an association. as above, they tried to account for many of the covariates for dementia (stroke, polypharmacy, depression, etc), and they also found that there was not much difference in use of the health care system (all suggesting, but not proving, that the PPI-users and non-users were pretty similar in dementia risk). But the data are not so granular: ie, even if the level of polypharmacy is the same in both groups, did those on PPIs might have more specific drugs that could affect cognition? and what about cigarette use, alcohol, or obesity which may be over-represented in the PPI-user group and may themselves be associated with dementia?

--so, i think there are a couple of lessons here:

    --it is always important to remember that very few drugs really are targeted specifically to a single action. one might think (as perhaps most of us did) that a drug targeting acid release from the stomach's parietal cells would not have potential widespread effects (other than collateral damage on nutrient absorption, etc). the finding of clear effects on b-amyloid in mice brains and the potential effects of PPIs on human brains yet again brings us to the usual conclusion: use medications in their lowest dose and only if necessary, with an emphasis on helping patients try to make lifestyle changes as the primary approach to many medical problems.

    --as mentioned in prior blogs, it is really easy to just keep refilling PPI prescriptions since they work and it is sometimes a struggle to change paths with patients, and we often are quite busy just keeping up with the other more active medical problems. ie, the approach of step-down therapy (moving from PPI to H2-blocker or just calcium) is in reality not done much in clinical practice. which i think supports more of a step-up approach (staring with calcium then escalating to H2 blocker prior to PPI).

for a few of the recent blogs on PPIs, see

http://gmodestmedblogs.blogspot.com/2014/11/gastric-acid-suppression-and-microbiome_23.html which looks at microbiome changes with gastric acid suppression

http://gmodestmedblogs.blogspot.com/2015/06/ppis-associated-with-mis_13.html/ which looks at the possible association of PPIs with MIs

http://gmodestmedblogs.blogspot.com/2016/01/ppis-and-chronic-kidney-dz.html found an association of PPIs with chronic kidney disease